You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The julia-runtest step has timeout-minutes: 5. Both macOS nightly jobs hit exactly that limit (312 s)
This is not just slow precompilation. After "gRPC server started", macOS sat silent for 3.5 minutes. Ubuntu nightly reached the docstring failures 78 s after server start, and before Sep 14 the whole macOS nightly suite finished in about 2 minutes. On this PR the docstring failures should have printed early on macOS too, but nothing printed. So it is stuck early: in grpc_init, the protojl codegen, or loading for @doc/REPL
It is not caused by your code. The last passing macOS nightly (run 34777430543, commit 8288f62) and the first failing one (main push 34883667653, e185ee2) have identical source trees. Every macOS nightly run since then has timed out the same way
The only thing that changed was Julia: DEV.3188 (f284acfe7bd, passing) became DEV.3194 (4d4de63063a, failing). Apart from two stdlib bumps (SparseArrays and Tar), there are two code commits in that range, and both touch loading and precompilation:
Activity
csvance commentedon Oct 1, 2026
It looks like its passing now...